Our comprehensive review [1] made two points. First, suggestions could be made about the selection of imaging tests (see Table 1 of accompanying position paper [2]). Second, the quality of published evidence limited our ability to make stronger recommendations. As a result, users of MRI are vulnerable to both underuse and overuse of a potentially excellent technology. None of those questioning our article actually challenge the College's clinical suggestions, and Dr. Baum's comments support the suggestions given for stroke.

The concerns expressed by the authors of the letters may reflect their particular practice backgrounds. Dr. Lossing cites individual cases in which MRI showed new disease. We suspect that other neurologists could cite cases with false-positive diagnoses and subsequent treatment misadventures caused by MRI results. The only way to avoid "dueling cases" is to cooperate and to do prospective studies as the standard for judging the effect and usefulness of MRI. Dr. Drayer's musings on horses and computers were eloquent, but his arguments are unconvincing without prospective data. Recently, a distressingly high rate of incidental findings has been reported in the back and the brain [3, 4], raising new concerns about potential false-positive interpretations.

Regarding the cost of prospective studies and the difficulty in finding patients with rare diseases, both are manageable, especially if specialty organizations contribute expertise and infrastructure to multi-institutional outcome registries. The deployment of radiologic advances into clinical practice should be based on solid data, not merely on salesmanship. Specialty physicians who use MRI must consider their situation carefully, given the complexity of the economics of self-interest that surrounds MRI [5].

Magnetic resonance imaging is an excellent and evolving diagnostic imaging technology. Costs are decreasing as MRI scanning techniques and designs improve. However, practicing internists need rigorous, objective information about the clinical efficacy of MRI (and all diagnostic tests) so they can apply these new technologies wisely. The raging storms over health care costs will throw lightning bolts at MRI unless its use is well grounded in prospective studies whose results fairly show when and where to use the technology to optimize patient outcomes.